Electronic Theses and Dissertations Theses, Dissertations, and Major Papers
12-20-2018
The Relationship between Fat Stereotypes and Body
The Relationship between Fat Stereotypes and Body
Dissatisfaction in Normal Weight Women: A Mediated Moderation
Dissatisfaction in Normal Weight Women: A Mediated Moderation
Model
Model
Jean Kim
University of Windsor
Follow this and additional works at: https://scholar.uwindsor.ca/etd
Recommended Citation Recommended Citation
Kim, Jean, "The Relationship between Fat Stereotypes and Body Dissatisfaction in Normal Weight Women: A Mediated Moderation Model" (2018). Electronic Theses and Dissertations. 7604.
https://scholar.uwindsor.ca/etd/7604
This online database contains the full-text of PhD dissertations and Masters’ theses of University of Windsor students from 1954 forward. These documents are made available for personal study and research purposes only, in accordance with the Canadian Copyright Act and the Creative Commons license—CC BY-NC-ND (Attribution, Non-Commercial, No Derivative Works). Under this license, works must always be attributed to the copyright holder (original author), cannot be used for any commercial purposes, and may not be altered. Any other use would require the permission of the copyright holder. Students may inquire about withdrawing their dissertation and/or thesis from this database. For additional inquiries, please contact the repository administrator via email
The Relationship between Fat Stereotypes and Body Dissatisfaction in Normal Weight Women: A Mediated Moderation Model
By
Jean Kim, M.A.
A Dissertation
Submitted to the Faculty of Graduate Studies through the Department of Psychology in Partial Fulfillment of the Requirements for
the Degree of Doctor of Philosophy at the University of Windsor
Windsor, Ontario, Canada
2018
The Relationship between Fat Stereotypes and Body Dissatisfaction in Normal Weight Women: A Mediated Moderated Model
by
Jean Kim
APPROVED BY:
______________________________________________ S. Spencer, External Examiner
Ohio State University
______________________________________________ K. Gorey
School of Social Work
______________________________________________ D. Jackson
Department of Psychology
______________________________________________ C.Y. Senn
Department of Psychology
______________________________________________ J.L. Jarry, Advisor
Department of Psychology
DECLARATION OF ORIGINALITY
I hereby certify that I am the sole author of this thesis and that no part of this thesis has been published or submitted for publication.
I certify that, to the best of my knowledge, my thesis does not infringe upon anyone’s copyright nor violate any proprietary rights and that any ideas, techniques, quotations, or any other material from the work of other people included in my thesis, published or otherwise, are fully acknowledged in accordance with the standard referencing practices. Furthermore, to the extent that I have included copyrighted material that surpasses the bounds of fair dealing within the meaning of the Canada Copyright Act, I certify that I have obtained a written permission from the copyright owner(s) to include such material(s) in my thesis and have included copies of such copyright clearances to my appendix.
ABSTRACT
Fat stereotypes refer to beliefs about traits that are considered characteristic of individuals carrying excess weight. Endorsing these beliefs is associated with negative body image in overweight and obese individuals. In normal weight women, however, these beliefs have a more nuanced effect on body image. The purpose of these studies was to extend
existing literature on the relationship between fat stereotype endorsement and body dissatisfaction in normal weight women. A mediated moderation model was proposed. Specifically, body surveillance was investigated as a moderator of the relationship between fat stereotype endorsement and body dissatisfaction. Further, downward
excess weight (e.g., genetics, food-rich environment, etc.) was presented. As
ACKNOWLEDGEMENTS
I am deeply appreciative of my supervisor, Dr. Josée Jarry, who had a
fundamental role in my journey to completing this doctorate. Her guidance, feedback, encouragement, and support throughout my time as her student were invaluable both to my professional and personal development. Thank you, Dr. Jarry, for your commitment to my growth over the past several years. I also would like to thank my committee members, Dr. Charlene Senn, Dr. Dennis Jackson, and Dr. Kevin Gorey. You offered thoughtful and conscientious feedback, encouraging me to consider new ideas that offered greater depth to my project. Thank you to Dr. Steven Spencer, who graciously agreed to be my external examiner, and who offered supportive and thought-provoking insights at the final stages of this project.
To my family, I am deeply grateful for your unwavering support. Mom, dad, uhn-ni, and Dave, your enduring belief in my abilities has offered me solace and stability in even the most challenging times. I have walked steadily through this journey, knowing that you were always there to listen, advise, and encourage. To my friends, thank you for your kindness, patience, and understanding throughout this process. Thank you for offering empathy and guidance during difficult stretches, allowing time to myself when needed, and then connecting with me to relax, laugh, and celebrate.
TABLE OF CONTENTS
DECLARATION OF ORIGINALITY ... iii
ABSTRACT ... iv
ACKNOWLEDGEMENTS ... vi
LIST OF TABLES ... xiii
LIST OF FIGURES ... xv
CHAPTER I. INTRODUCTION ... 1
Clinical Relevance of Endorsing Fat Stereotypes ... 1
Body dissatisfaction and fat stereotypes ... 2
Body Surveillance ... 5
Body surveillance and body dissatisfaction ... 7
Body surveillance and internalized thin ideals ... 7
Social Comparison Theory ... 9
Physical appearance comparison and body dissatisfaction ... 11
Physical appearance comparison and body surveillance ... 13
Proposed Model of Associations between Body Dissatisfaction, Fat Stereotypes, Body Surveillance, and Downward Physical Appearance Comparison ... 15
Higher fat stereotypes and higher body surveillance ... 15
Lower fat stereotypes and higher body surveillance ... 18
Higher fat stereotypes and lower body surveillance ... 18
Lower fat stereotypes and lower body surveillance ... 19
Overview of Two Studies ... 23
II. STUDY 1 ... 25
Purpose and Hypotheses ... 25
METHOD ... 26
Participants ... 26
Measures ... 28
Predictor variable: Obese Persons Trait Survey ... 28
Moderator variable: Objectified Body Consciousness Surveillance Subscale ... 29
Mediator variable: Downward Physical Appearance Comparison Scale ... 29
Criterion variable: Eating Disorder Inventory-2 - Body Dissatisfaction ... 30
Covariates ... 30
Procedure ... 32
RESULTS ... 33
Approach to Data Analysis ... 33
Main Analyses for Caucasian-Only Sample ... 33
Data preparation ... 33
Assumptions of multiple regression ... 34
Moderation analysis for body dissatisfaction ... 38
Mediated moderation analysis ... 44
Data preparation ... 49
Assumptions of multiple regression ... 49
Moderation analysis for body dissatisfaction ... 50
Mediated moderation analysis for full sample ... 55
DISCUSSION ... 58
III. STUDY 2 ... 62
Purposes and Hypotheses ... 66
METHOD ... 68
Design ... 68
Participants ... 69
Materials ... 71
Measures ... 72
Moderator variable: Objectified Body Consciousness Surveillance Subscale ... 72
Mediator variable: State Downward Physical Appearance Comparison Scale ... 72
Criterion variables: Body Image States Scale ... 72
Figure Rating Scale ... 73
Covariates ... 73
Manipulation check: Obese Persons Trait Survey ... 74
Procedure ... 74
Approach to Data Analysis ... 78
Main Analyses for Caucasian-Only Sample ... 79
Data preparation ... 79
Assumptions of multiple regression ... 79
Moderation analysis for state body dissatisfaction ... 83
Mediated moderation analysis ... 89
Moderation analysis for figure rating discrepancy ... 93
Analyses for Full Ethnically Heterogeneous Sample ... 96
Data preparation ... 96
Assumptions of multiple regression ... 96
Moderation analysis for state body dissatisfaction ... 97
Mediated moderation analysis for ethnically heterogeneous sample ... 100
Moderation analysisfor figure rating discrepancy in ethnically heterogeneous sample ... 105
DISCUSSION ... 108
Downward Physical Appearance Comparison as an Explanatory Mechanism ... 109
Possible Role of Locus of Control? ... 110
Figure Rating Discrepancies as Indicator of Contrast vs. Assimilation Effects ... 112
IV. GENERAL DISCUSSION ... 114
Lower Body Surveillance: Activation of State Locus of Control ... 117
Influence of Race and Ethnicity ... 119
Clinical and Social Applications ... 121
Limitations and Future Directions ... 125
Sampling limitations ... 125
Fat stereotype endorsement versus anti-fat attitudes ... 128
Fat stereotype manipulation and locus of control measurement 128 Other effects of fat stereotypes? ... 129
State downward physical appearance comparison ... 129
Conclusion ... 130
REFERENCES ... 132
APPENDICES ... 150
Appendix A Obese Persons Trait Survey ... 150
Appendix B Objectified Body Consciousness Scale Body Surveillance Subscale ... 156
Appendix C Upward and Downward Physical Appearance Comparison Scale ... 158
Appendix D Eating Disorder Inventory-2 – Body Dissatisfaction Subscale ... 160
Appendix E Sociocultural Attitudes Towards Appearance Scale-3 ... 161
Appendix F Beck Depression Inventory-II ... 163
Appendix G Marlowe-Crowne Social Desirability Scale-Form C ... 165
Appendix H Rosenberg Self-Esteem Scale ... 166
Appendix I Demographic Questionnaire ... 167
Appendix J Participant Pool Recruitment Advertisement – Study 1 ... 168
Appendix L Post-Study Information Debriefing form ... 171
Appendix M Mock Health Report: Support Condition ... 173
Appendix N Mock Health Report: Challenge Condition ... 176
Appendix O State Upward and Downward Physical Appearance Comparison Scale .... 179
Appendix P Body Image States Scale ... 181
Appendix Q Figure Rating Scale ... 184
Appendix R Participant Pool Recruitment Advertisement – Study 2 ... 185
Appendix S Letter of Information for Consent to Participate in Research ... 186
Appendix T Post-Study Information Debriefing Form ... 188
Appendix U Letter of Information for Consent to Participate in Research ... 190
Appendix V Brief Memory Test ... 192
Appendix W Information and Debriefing Form ... 194
LIST OF TABLES STUDY 1
Table 1. Zero-Order Correlations Between All Variables for Caucasian Sample ... 36
Table 2. Descriptive Statistics for Caucasian Sample ... 39
Table 3. Moderation Model Summary Predicting Body Dissatisfaction for Caucasian Sample ... 41
Table 4. Effects of Fat Stereotypes on Body Dissatisfaction at Varying Levels of Body Surveillance for Caucasian Sample ... 42
Table 5. Mediated Moderation Model Summaryfor Caucasian Sample ... 48
Table 6. Zero-Order Correlations Between All Variables for Full Sample ... 51
Table 7. Descriptive Statistics for Full Sample ... 52
Table 8. Moderation Model Summary Predicting Body Dissatisfaction for Full Sample ... 54
Table 9. Mediated Moderation Model Summary for Full Sample ... 57
STUDY 2 Table 10. Zero-Order Correlations Between All Variables for Caucasian Sample .. 82
Table 11. Descriptive Statistics by Condition for Caucasian Sample ... 84
Table 12. Moderation Model Summary Predicting State Body Dissatisfaction for Caucasian Sample ... 86
Table 13. Effects of Study Condition on State Body Dissatisfaction at Varying Levels of Body Surveillance for Caucasian Sample ... 87
Table 15. Moderation Model Predicting Figure Rating Discrepancies for Caucasian Sample ... 95 Table 16. Zero-Order Correlations Between All Variables for Full Sample ... 98 Table 17. Descriptive Statistics by Condition for Full Sample ... 99 Table 18. Moderation Model Summary Predicting State Body Dissatisfaction for
Full Sample ... 101 Table 19. Mediated Moderation Model Summary for Full Sample ... 104 Table 20. Moderation Model Summary Predicting Figure Rating Discrepancies for
LIST OF FIGURES STUDY 1
Figure 1. Proposed conceptual mediated moderation model ... 17 Figure 2. The relationship between fat stereotype endorsement and body
dissatisfaction at lower and higher levels of body surveillance ... 43 Figure 3. Statistical diagram for predicted mediated moderation model for Study 1
... 45 Figure 4. Statistical diagram for predicted mediated moderation model for full
sample in Study 1 ... 56 STUDY 2
Figure 5. The impact of condition on state body dissatisfaction at lower and higher levels of body surveillance ... 88 Figure 6. Statistical diagram for predicted mediated moderation model for Study 2
... 90 Figure 7. Statistical diagram for predicted mediated moderation model for full
Chapter I
The Relationship between Fat Stereotypes and Body Dissatisfaction in Normal Weight Women: A Mediated Moderation Model
Increasing societal focus on health and physical appearance has generated
concerns about the stigmatization of overweight and obese individuals. Prejudice against this group is described as one of the last acceptable forms of bias in modern society (Puhl & Brownell, 2001). Indeed, research indicates that negative attitudes toward overweight and obese individuals are considered to be more acceptable than are negative attitudes toward other groups, including individuals with physical disabilities (Latner, Stunkard, & Wilson, 2005), with AIDS, and those of various races (Crandall, Eshleman, & O’Brien, 2002). Further, weight-based prejudice and discrimination have been documented across many domains of life. Overweight and obese individuals receive unfair treatment in employment and hiring (e.g., Roehling, 1999), health care services (e.g., Brochu & Esses, 2009; Schwartz, Chambliss, Brownell, Blair, & Billington, 2003), and education (e.g., Crandall, 1991; Puhl & Latner, 2007). Given the pervasiveness of anti-fat stigma, researchers have investigated several social and clinical consequences that result from weight bias (see Puhl & Heuer, 2009 for a review). Much of the focus of research has been on how fat stereotyping affects its recipients. The current research aims to extend the literature examining how fat stereotypes affect those who hold them.
Clinical Relevance of Endorsing Fat Stereotypes
unintelligent, and lack willpower (Puhl & Brownell, 2001), among others. Numerous studies have shown that people of all weight categories commonly endorse these fat stereotypes. For example, a population-based investigation found that 23.5% of 1000 participants held “definite stigmatizing attitudes” toward obese individuals, with no difference in reported levels of stigma across genders (Hilbert, Rief, & Braehler, 2008). Stigmatizing attitudes were defined in this study as the extent of agreement with
statements exemplifying fat stereotypes, such as “fat people have no willpower” and “most fat people are lazy”. Similarly, Swami, Pietschnig, Stieger, Tovée, and Voracek (2010) examined the extent to which individuals endorse fat stereotypes. These authors found that across their total sample of 1024 participants, fat stereotypes such as laziness and insecurity were moderately endorsed. These large-scale studies indicate that fat stereotypes are commonly held in modern society.
Given the research demonstrating that fat stereotypes are commonly held, researchers have investigated potential negative mental health outcomes resulting from this endorsement. This research has focussed mostly on negative outcomes in overweight and obese individuals who themselves endorse fat stereotypes. Generally, these studies indicate that holding fat stereotypes is associated with a number of negative
psychological outcomes, such as low self-esteem (Friedman et al., 2005; Klaczynski, Goold, & Mudry, 2004), depressive symptomatology (Durso & Latner, 2008; Friedman et al., 2005), anxiety (Durso & Latner, 2008) and, notably, body dissatisfaction
(Friedman et al., 2005).
evaluation of one’s body, including body shape, weight, and specific body parts, such as one’s stomach or thighs (Stice & Shaw, 2002). Body dissatisfaction is considered to be a key predictor of the development of disordered eating behaviours (Stice, 2001), and is described as an essential precursor to clinical eating disorders (Polivy & Herman, 2002). This demonstrated relationship between body dissatisfaction and eating disorders is one reason why researchers have investigated the factors that contribute to the development of body dissatisfaction. Further, though body dissatisfaction commonly is reported among men and women of all weight categories, women tend to report greater body
dissatisfaction than do men (Cash, Morrow, Hrabosky, & Perry, 2004; Frederick, Forbes, Grigorian, & Jarcho, 2007). Indeed, body dissatisfaction is so widespread amongst North American girls and women that it has been considered normative for over three decades (Rodin, Silberstein, & Striegel-Moore, 1984). Additionally, women tend to report greater overweight preoccupation than do men (Cash et al., 2004), and women constitute the vast majority of eating disorders sufferers (Ingram & Price, 2010). Therefore, understanding the factors contributing to body dissatisfaction, especially in women, is of prime
importance. Accordingly, the current research focuses on the potential impact of fat stereotype endorsement on body dissatisfaction in women.
Research on the relationship between body dissatisfaction and endorsed fat stereotypes is fairly recent and disproportionately conducted with overweight and obese samples. Findings vary slightly depending on whether the measurement of fat stereotypes is implicit or explicit. Implicit measures indirectly assess automatic beliefs and are
Despite some variability across measurement methods, findings generally indicate that overweight and obese individuals who endorse fat stereotypes report greater body dissatisfaction than do overweight and obese individuals who do not endorse these stereotypes. For example, Friedman et al. (2005) found that obese men and women who explicitly endorsed negative fat stereotypes reported greater body image distress than did those who did not endorse these stereotypes. Similarly, Durso and Latner (2008) found that greater explicit endorsement of fat stereotypes was related to greater body shape concerns in overweight and obese men and women. Further, Carels et al. (2010) found that greater implicit weight bias was related to lower body satisfaction in overweight and obese men and women.
The foregoing evidence indicates that endorsing fat stereotypes is associated with body dissatisfaction in overweight and obese individuals. However, few studies have investigated the association between fat stereotype endorsement and body dissatisfaction in average weight women. Further, investigations of this relationship tend to assess the specific fat stereotypes of willpower and controllability, which refer to the notion that one’s weight is a matter of willpower and self-control. For example, Laliberte, Newton, McCabe, and Mills (2007) found that endorsing the belief that weight is completely controllable was related to higher body dissatisfaction in a predominantly normal weight sample of women. Participants with lower endorsement of this belief tended to report lower body dissatisfaction. Similarly, O’Brien, Hunter, Halberstadt, and Anderson (2007) found, in a predominantly normal weight sample of men and women, that greater
predominantly normal weight male and female physical education students with higher levels of implicit fat stereotypes reported significantly greater body dissatisfaction than did those with lower levels of implicit fat stereotypes. These same students also reported greater explicit endorsement of the willpower stereotype. It has been suggested that individuals who strongly believe that weight is completely controllable and is a matter of willpower likely feel dissatisfied with their own body because they feel responsible for their failure at maintaining an ideal weight (Laliberte et al., 2007). Because these ideals often are unattainable or difficult to maintain, belief in such stereotypes is thought to engender body dissatisfaction, even in normal weight women.
To expand upon this limited research base, a recent study by Kim and Jarry (2014) examined the relationship between fat stereotypes and body dissatisfaction in a sample of normal weight Caucasian women. In contrast to the studies described above, there was no significant relationship between fat stereotypes and body dissatisfaction in this sample (r = .06). One notable difference in the Kim and Jarry (2014) study was the use of a measure assessing general fat stereotypes (e.g., laziness, uncleanliness,
unintelligence, lack of willpower), rather than a measure exclusively related to appearance-based stereotypes (e.g., unattractive). Thus, it appears that beliefs about general negative traits associated with overweight and obese individuals are not directly associated with body dissatisfaction in normal weight women. However, Kim and Jarry (2014) reported a subtler role for fat stereotypes in the body image of these women, discussed below.
Body Surveillance
stereotypes and body image, little is known of the factors that impact this potential association. Kim and Jarry (2014), however, investigated body surveillance as a potential vulnerability factor. Though they did not find a direct relationship between fat stereotype endorsement and body dissatisfaction in their sample, the authors reported a moderating effect of body surveillance. Body surveillance refers to the tendency to view one’s body from the perspective of an outside observer (Fredrickson & Roberts, 1997; McKinley & Hyde, 1996). In accordance with this definition, women with high body surveillance are those who report a greater tendency to look at and monitor their body frequently
(McKinley, 1998; McKinley & Hyde, 1996). Further, women with high body
surveillance are greatly concerned with how their body looks rather than how it feels (McKinley, 1998).
Gender differences in body surveillance have been examined. Though body surveillance is reported by both women and men, women consistently report higher levels than do men (e.g., Frederick et al., 2007; Lowery et al., 2005; McKinley, 1998). Further, women continue to show greater levels of body surveillance than do men as they age (McKinley, 2006). Notably, studies that consistently demonstrate moderate to high levels of body surveillance in women tend to use samples with an average body mass index (BMI) classified as normal weight (e.g., Brannan & Petrie, 2008; Fitzsimmons & Bardone-Cone, 2011; Greenleaf & McGreer, 2006; Sinclair & Myers, 2004). Further, Mercurio and Rima (2011) reported no relationship between BMI and body surveillance in a sample of undergraduate women. Accordingly, this research indicates that not only do women engage in body surveillance more frequently than do men, they do so
Body surveillance and body dissatisfaction. Kim and Jarry (2014) assessed body surveillance as a moderating factor in their study in part due to its documented relationship with body dissatisfaction. In general, the tendency to look at and monitor one’s body frequently is thought to be harmful to body satisfaction in women. Body surveillance is theorized to lead to increased body dissatisfaction because it promotes an awareness of the discrepancy between one’s own body and internalized cultural standards of attractiveness (McKinley & Hyde, 1996). Indeed, Brannan and Petrie (2008), as well as Mercurio and Rima (2011), observed a strong positive relationship between body dissatisfaction and body surveillance in women. Further, this relationship has been documented across different BMI categories. For example, Frederick et al. (2007) found that higher body surveillance was related to higher body dissatisfaction in normal weight, overweight, and obese women. Though the relationship was more pronounced in
overweight and obese women, this finding supports that normal weight women who habitually monitor their body also tend to experience greater body dissatisfaction.
Body surveillance and internalized thin ideals. Researchers have investigated how the internalization of cultural body standards is related to body surveillance
(McKinley & Hyde, 1996). In particular, this literature has focussed on internalized thin ideals, which refers to the extent to which an individual endorses societal portrayals of thinness as being the epitome of beauty (Thompson & Stice, 2001). For example,
Fitzsimmons-Craft et al. (2012) investigated the relationships between body surveillance, internalized thin ideals, and body dissatisfaction in undergraduate women. Body
surveillance was positively related both to internalized thin ideals and to body
between body surveillance and internalized thin ideals, and between body surveillance and body dissatisfaction, in their sample of normal weight undergraduate women.
As described, the literature on body surveillance has focussed on assessing its relationship with cultural ideals of thinness. Given the prevalence of fat stigma outlined above, however, it is possible that body surveillance plays a role in body dissatisfaction not only in individuals who internalize the ideal of thinness, but also in individuals who endorse the negative connotations characterized in fat stereotypes. Kim and Jarry (2014) argued that “body surveillance may further promote body dissatisfaction in women who endorse fat stereotypes by heightening their awareness of their own body fat, a
characteristic that they denigrate” (p. 332). This seems especially likely, given that the desire to avoid being overweight may be more strongly related to negative body image than is the desire to achieve thinness (e.g., Dalley & Buunk, 2009; Woud, Anschutz, Van Strien, & Becker, 2011). Thus, Kim and Jarry (2014) suggested that women who
habitually monitor their body and who also denigrate fat might be more susceptible to body dissatisfaction. Their close body monitoring may heighten awareness of unwanted fat on their body, especially if the latter is imbued with the negative meaning associated with fat stereotypes.
appeared to serve a defensive function against body dissatisfaction in normal weight women who habitually monitored their body. Further, in women reporting lower levels of body surveillance, greater fat stereotype endorsement was related to higher body
dissatisfaction. Thus, for women who monitored their body less frequently, strongly endorsing fat stereotypes appeared to be detrimental to their body image. The authors purported that social comparison, specifically physical appearance comparison, could explain this paradoxical finding.
Social Comparison Theory
The Social Comparison Theory states that people compare themselves to others in order to evaluate the self (Festinger, 1954). It also posits that people have an innate drive to evaluate various dimensions of the self (e.g., skills, attitudes, status). When objective means for self-evaluation are not possible, people will compare themselves to others to develop these evaluative judgments of the self (Festinger, 1954). Though it is considered a natural process in all humans, frequent engagement in social comparison tends to be related to negative factors across several domains. For example, a greater tendency to compare oneself to others is related to lower self-esteem, greater social anxiety, greater self-consciousness, and a greater tendency to engage in negative behaviours such as lying (Gibbons & Buunk, 1999).
less fortunate other, termed downward comparison, can increase one’s own subjective well-being. Wills (1981) also proposed that “downward comparison can be achieved through active derogation of another person, thereby increasing the psychological distance between the self and the [inferior] other” (p. 246). Conversely, comparing oneself to more fortunate others, termed upward comparisons, can decrease subjective well-being when the comparison promotes the contrast between oneself and the superior other (Wheeler, 1966; Collins, 1996).
mutable, however, are thought to produce larger assimilation effects.
Applying these concepts to directional social comparison processes allows for more specific hypotheses about their resulting effects on subjective well-being. Specifically, downward social comparisons in which the target is distinct and the self-view is clear are likely to result in greater well-being through a contrast effect that promotes the differences between the self and the inferior target. Downward social comparisons in which the target is indistinct and the self-view is mutable, however, are more likely to reduce well-being through an assimilation effect that promotes the similarities between the self and the inferior target. The opposite consequences for well-being are theorized for upward social comparisons. Upward social comparisons in which the target is distinct and the self-view is clear are likely to reduce well-being through a contrast effect that promotes the differences between the self and the superior target. Upward social comparisons in which the target is indistinct and the self-view is mutable are likely to increase well-being through an assimilation effect that promotes the
similarities, or the possibility of similarities, between the self and the superior targets. These principles of social comparison have been applied in the area of body image, referred to as physical appearance comparison (Thompson, Heinberg, & Tantleff, 1991).
Thompson et al., 1991). Myers and Crowther (2009) conducted a meta-analysis of 156 studies (189 effect sizes) and showed that greater engagement in physical appearance comparisons is related to higher body dissatisfaction.
Though physical appearance comparison has been assessed as a unitary concept in much of the literature, the “downward” and “upward” concepts recently have been
applied to appearance-based comparisons. Specifically, downward appearance
comparison has been described as comparing oneself to people who are perceived as less attractive, often including individuals considered overweight or obese (O’Brien et al., 2009). In contrast, upward appearance comparison involves comparing oneself to people perceived as more attractive, often including individuals who are thinner (O’Brien et al., 2009). As described by O’Brien et al. (2009), the underlying assumption of the physical appearance comparison research is that people tend to make upward physical appearance comparisons rather than downward comparisons, which leads to greater body discontent. In support of this notion, some research has shown that college women engage in more upward than downward physical appearance comparison (Tiggeman & Polivy, 2010). However, O’Brien et al. (2009) argue that given the different outcomes of upward and downward social comparison on well-being, it is likely that upward and downward physical appearance comparison have distinct effects on body image.
comparison was related to more positive body image. Further, greater engagement in upward appearance comparison was related to more negative body image. Additionally, greater downward physical appearance comparison, but not upward physical appearance comparison, was related to stronger anti-fat attitudes (O’Brien et al., 2009). Similarly, Bailey and Ricciardelli (2010) found that more upward appearance comparisons and less downward appearance comparisons predicted higher body dissatisfaction and higher eating disturbance. Further, Leahey, Crowther, and Mickelson (2007) found that exposure to less attractive others, which likely leads to downward comparison, induced more positive self-evaluations. Interestingly, however, Vartanian and Dey (2013) found that both greater downward and upward physical appearance comparison were related to greater body dissatisfaction in undergraduate women. Further, these authors found that a weaker and unstable sense of self was related to greater engagement in both downward and upward physical appearance comparison. This points to the complex influence of physical appearance comparison on body satisfaction, suggesting that unidirectional comparisons may differentially influence body appraisals depending on other factors. In general, however, it is theorized that downward physical appearance is related to
improvements in body satisfaction.
surveillance positively predicted the tendency to engage in body-based comparisons. Similarly, Fitzsimmons-Craft et al. (2012) examined this relationship in undergraduate women. In accordance with the above study, these researchers found a significant positive relationship between body surveillance and the tendency to engage in physical appearance comparisons. These findings support the notion that higher body surveillance is associated with higher appearance comparison. One limitation noted in both studies, however, is that the measure of physical appearance comparison did not distinguish between upward or downward comparisons, and instead assessed physical appearance comparisons in general. To date, researchers have theorized that upward physical appearance comparison reminds women who have a greater tendency to monitor their body that they fall short of the internalized thin ideal. However, differential assessment of upward and downward comparisons is not commonly conducted in these studies. Thus, it is difficult to determine whether one or both directions of comparisons are related to body surveillance. Though it is highly likely that normal weight women who monitor their body frequently compare themselves to targets perceived to be more attractive, it also is possible that they compare themselves to targets perceived to be less attractive to improve their subjective appraisal of their appearance. This might be especially likely if they hold negative evaluations of overweight and obese individuals, a group often perceived as unattractive.
satisfaction (Bailey & Ricciardelli, 2010), though there are some discrepant findings for the latter (Vartanian & Dey, 2013). Thus, in women with greater body surveillance, a tendency to hold negative beliefs about overweight individuals might lead to the
engagement of downward physical appearance comparison with this target group, serving to improve body dissatisfaction. Further, the combination of varying levels of fat
stereotypes and body surveillance may differentially influence downward comparison and, subsequently, body dissatisfaction, as described below.
Proposed Model of Associations between Body Dissatisfaction, Fat Stereotypes,
Body Surveillance, and Downward Physical Appearance Comparison
Based on the reviewed literature, the unexpected effect reported by Kim and Jarry (2014) could be explained through differential downward physical appearance
comparisons. The proposed conceptual mediated moderation model is presented in Figure 1. In this model, it was predicted that body surveillance moderates the relationship
between fat stereotypes and body dissatisfaction, and that this moderation is mediated by downward physical appearance comparison. The theoretical explanations of the proposed relationships are as follows.
Higher fat stereotypes and higher body surveillance. Normal weight women with higher body surveillance and who also strongly endorse fat stereotypes may be most likely to engage in downward appearance comparisons against larger people perceived to be less attractive. Using Stapel and Koomen’s (2000) social comparison concepts
Figure 1. Proposed conceptual mediated moderation model. Fat Stereotypes
Downward Physical Appearance Comparison
Lower fat stereotypes and higher body surveillance. In contrast, normal weight women who have a higher tendency to monitor their body but who do not strongly
endorse fat stereotypes may be less likely to engage in downward appearance
comparisons with overweight and obese individuals because of their lower derogation of this group. Though they may have clear and immutable self-views, the targets of
comparison may be less distinct from the self because of their lower fat stereotypes, creating less favourable conditions for a pronounced contrast effect. Their lower level of downward comparison may result in less self-enhancement due to the weaker salience of the discrepancy between themselves and obese individuals, hence their higher reported body dissatisfaction. Therefore, women who report higher body surveillance and who hold less fat stereotypes may not be protected by the contrast effects of downward comparison. Further, in comparison to women with high surveillance and high fat
stereotypes, assimilation effects rather than contrast effects may occur when these women do engage in downward comparison due to the indistinct boundaries between themselves and the comparison targets. Finally, these women may be acutely focussed on aspects of their body with which they are dissatisfied. The combination of less downward
appearance comparison with large people and the internalization of a thin ideal that they are unlikely to match may create conditions favourable to the development of body dissatisfaction in this group of women.
lower focus on their body, these women may be less likely to engage in downward physical appearance in comparison to women with high body surveillance (Tylka & Sabik, 2010). Because of their fat stereotypes, however, they are more likely to engage in downward comparison than are women with low surveillance and low fat stereotypes. Further, because these women do not tend to monitor and examine their body frequently, they may have mutable and vague self-views, creating conditions favourable for an assimilation effect rather than a contrast effect when they do engage in downward comparison. As such, the discrepancy between themselves and obese individuals may be less salient to them, and they may actually focus on the similarities between themselves and the target group that they denigrate. This would generate greater dissatisfaction with their own body. It is notable, however, that these women still report lower body
dissatisfaction than do women who engage in higher body surveillance, regardless of the latter’s degree of fat stereotype endorsement.
The current research sought to integrate these theoretical explanations into a proposed mediated moderation model. It was expected that downward physical appearance comparison in particular would be the mechanism through which the interaction between fat stereotype endorsement and body surveillance influences body dissatisfaction. Because upward physical appearance comparison against obese targets is unexpected (O’Brien et al., 2009), the proposed research did not examine the impact of this directional comparison as a mediating variable. Based on the theoretical explanations outlined above, it was expected that this interaction between fat stereotyping and body surveillance would influence one’s tendency to engage in downward appearance
comparison, which in turn, would impact body dissatisfaction. In summary, women with higher body surveillance and higher fat stereotypes were predicted to engage most extensively in downward comparison, with a resulting contrast effect leading to lower body dissatisfaction. Women with lower body surveillance were expected to engage in relatively less downward comparison, but with a resulting assimilation effect when they also endorsed high fat stereotypes. Finally, it was expected that the overall effect of downward comparison on body dissatisfaction would be most influenced by women with higher body surveillance and higher fat stereotypes, indicating a negative relationship between these two variables.
Influence of Race and Ethnicity on Body Image and Weight-Bias
Caucasian sample of undergraduate women, it was not significant in the ethnically heterogeneous sample. The research on the influence of race and ethnicity on weight bias and body image is summarized below.
individuals.
In addition to weight bias, racial differences in body dissatisfaction have been documented. For example, Akan and Grilo (1995) assessed body dissatisfaction in African American, Asian American, and Caucasian undergraduate students. Caucasian participants reported greater levels of body dissatisfaction than did both Asian American and African American participants. In contrast, Asian American and African American participants reported similar levels of body dissatisfaction. Interestingly, a history of weight-related teasing was related to body dissatisfaction both in African American and Caucasian American participants, but not in Asian American participants. This suggests that even across races reporting similar levels of body satisfaction, such as Asian American and African American, differences in the relationships between body
satisfaction and weight-related constructs exist. Indeed, the results of the Kim and Jarry (2014) study showed that body surveillance moderated the relationship between fat stereotype endorsement and body dissatisfaction in Caucasian normal weight women only. This association disappeared when non-Caucasian women were integrated in the sample for analyses. Further, a meta-analysis found that across 98 studies, Caucasian women tend to report greater body dissatisfaction than do Hispanic and Black women, and that Hispanic women tend to report greater body dissatisfaction than do Black women (Grabe & Hyde, 2006). Additionally, this meta-analysis found that Asian women do not report significantly different levels of body dissatisfaction compared to Black, Caucasian, or Hispanic women (Grabe & Hyde, 2006).
However, their findings supported the notion that race continues to play an important role in body image research. To follow up on these findings, the main analyses of the current research were conducted on Caucasian normal weight women. Data also were collected from participants of other races and ethnicities to conduct additional analyses examining the same mediated moderation model in an ethnically heterogeneous sample.
Overview of Two Studies
The goal of this research was to extend the knowledge on the impact of fat stereotypes and body surveillance in normal weight women. Using two studies, the research built upon Kim and Jarry’s (2014) findings by examining downward physical appearance comparison as the mechanism through which the interaction between fat stereotype endorsement and body surveillance influences body dissatisfaction. The proposed research also examined the causal role of fat stereotype endorsement in the mediated moderation model.
The first study utilized self-report measures of fat stereotype endorsement, body surveillance, downward physical appearance comparison, and body dissatisfaction to test the proposed mediated moderation model. In addition to these variables, BMI (Frederick et al., 2007), global self-esteem (Lowery et al., 2005), and depressive symptoms
surveillance (Fitzsimmons-Craft et al., 2012), thin ideal internalization was controlled in the analyses. This ensured that the proposed mediated moderation model was examined above and beyond the critical effect of the thin ideal.
In the second study, an experimental design was used to examine the same proposed model. The model suggests that fat stereotyping has a potential causal role in the experience of body dissatisfaction, albeit indirectly. If holding fat stereotypes truly protects against body dissatisfaction in normal weight women with higher body surveillance, experimentally increasing endorsement of fat stereotypes should result in lower body dissatisfaction. Further, experimentally increasing endorsement was expected to result in higher body dissatisfaction in women with lower body surveillance, given the explanations described above. In this study, participants first completed a demographic questionnaire and measures of body surveillance, internalized thin ideals, trait
self-esteem, depressive symptoms, and socially desirable responding. As in the first study, the latter three measures were assessed as potential covariates. Participants then were
randomly assigned either into a condition intended to increase fat stereotype endorsement by presenting information that supports these stereotypes (support condition), or to a condition intended to decrease fat stereotype endorsement by presenting information that challenges these stereotypes (challenge condition). These conditions are described in detail below. Then, participants completed measures of state body dissatisfaction, state downward physical appearance comparison, and endorsement of fat stereotypes.
and the resulting impact on attitudes and discriminatory behaviour against overweight and obese individuals (Daníelsdóttir, O'Brien, & Ciao, 2010). These studies, however, have not investigated the impact of weight bias reduction on appraisals of one’s body, or on the tendency to compare one’s body to that of overweight and obese individuals. If the latter were found, this would be an additional potential benefit of weight-bias reduction programmes by decreasing the likelihood of women disparaging others to increase their own body satisfaction. The details of the two studies are presented below.
Chapter II
Study 1
Purpose and Hypotheses
The first purpose of Study 1 was to replicate the moderation effect reported by Kim and Jarry (2014). Body surveillance was predicted to moderate the relationship between fat stereotype endorsement and body dissatisfaction. The second purpose of Study 1 was to extend the findings reported by Kim and Jarry (2014) by examining downward physical appearance comparison as an explanation for this moderated effect. In other words, downward physical appearance comparison was expected to be the mechanism through which the interaction between fat stereotype endorsement and body surveillance impacts body dissatisfaction. The specific hypotheses for Study 1 are outlined below:
negative relationship). Conversely, for normal weight women with lower levels of body surveillance, higher fat stereotype endorsement will be related to higher body dissatisfaction (i.e., a positive relationship).
2. This moderated effect will be mediated by downward physical appearance comparison. Specifically, the interaction between fat stereotypes and body surveillance will predict downward physical appearance comparison, such that at higher levels of body surveillance, greater fat stereotype endorsement will be related to greater downward comparison. At lower levels of body surveillance, greater fat stereotypes also will be related to greater downward comparison, but to a lesser degree than in the high surveillance women. Finally, greater downward comparison is expected to predict lower body dissatisfaction while controlling for the interaction between fat stereotypes and body surveillance, thus completing the mediated moderation model.
Study 1: Method
Participants
Participants were recruited from the Psychology Participant Pool at the University of Windsor and received 0.5% course credit for their participation. Because the
calculations based on this self-reported information are highly specific in identifying people of normal weight (e.g., Larsen et al. 2008; Brener et al., 2003). Notably, Larsen et al. (2008) reported that 98.9% of normal weight women were correctly identified based on self-reported weight and height. Further, Larsen et al. (2008) reported a correlation of r = .94 between self-reported and objective BMI. Finally, past studies consistently indicate that inaccuracies in self-reported weight information are a greater concern at higher BMIs (e.g., Larsen et al., 2008; Cash et al., 1992; McCabe et al., 2001). Based on this information, it was assumed that self-reported weight and height information could be used reliably to calculate BMI. Analyses were conducted only on participants with a BMI between 18.5 to 25 kg/m2. Additionally, main analyses were conducted on
Caucasian participants only, though additional analyses were conducted on the full, ethnically heterogeneous sample.
Data were collected from a total of 654 participants. Of these participants, 404 were normal weight. The mean age of participants was 19.91 years (SD = 2.59) and their mean self-reported BMI was 21.65 kg/m2 (SD = 1.67). Self-reported race and ethnicity were as follows: 76.0% Caucasian (n = 307), 7.7% Arab or West Asian, 6.7% South Asian, 4.0% African Canadian, 3.2% East Asian, 0.5% South American, 0.2% Native Canadian, and 1.8% reported two or more ethnic backgrounds. Further, 93.6% reported no lifetime diagnosis of an eating disorder, 5.9% reported having been diagnosed previously, and 0.5% did not report if they had ever been diagnosed with an eating disorder.
and 4.7% had attended university for more than four years. Additionally, 49.1% of participants were psychology majors. In terms of current employment status, 65.3% were employed part-time, 32.4% were unemployed, and 2% were employed full-time.
Measures
Predictor variable. The Obese Persons Trait Survey (OPTS; Puhl, Schwartz, & Brownell, 2005; Appendix A) is a 20-item self-report measure that assesses endorsement of traits associated with obese persons. The OPTS consists of two subscales. The
OPTSneg measures endorsement of negative stereotypes and lists 10 negative traits, such as laziness. The OPTSpos measures positive stereotypes and lists 10 positive traits, such as generous. Participants are asked to estimate the percentage (0-100%) of obese persons who possess each of these traits. Although the entire scale was administered in this study to maintain psychometric properties, only the OPTSneg subscale was used in the
analyses. Based on the method described by Carels et al. (2010), participants also were asked to estimate the percentage of average weight persons who possess the same 20 traits. The Average-Weight Persons Trait survey (APTS) consists of both the negative (APTSneg) and positive (APTSpos) subscales, but only the APTSneg was used in the analyses. To obtain an indicator of the extent to which participants endorsed fat
2012; Puhl et al., 2005). The APTSneg also has demonstrated excellent internal consistency, α = .90 (Carels et al., 2010). In the current study, the OPTSneg and APTSneg both had good internal consistency, with α = .84 and α = .84, respectively.
Moderator variable. The Objectified Body Consciousness Surveillance Subscale (OBCSS; McKinley & Hyde, 1996; Appendix B) is an 8-item self-report measure that assesses the tendency to engage in body surveillance, or to closely examine one’s body. Participants respond on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). A sample item is “During the day, I think about how I look many times.” Higher scores indicate greater body surveillance. Internal consistencies for the OBCSS have ranged from .81 to .89 in past research (Brannan & Petrie, 2008; McKinley & Hyde, 1996). This subscale also has demonstrated convergent validity with a measure of appearance orientation (r = .64; McKinley & Hyde, 1996). In the current study, the OBCSS had good internal consistency, α = .84.
Mediator variable. The Downward Physical Appearance Comparison (DPACS) (O’Brien et al., 2009; Appendix C) is an 8-item self-report measure that assesses the tendency to compare oneself with targets perceived as less physically attractive.
Criterion variable. The Eating Disorder Inventory-2 Body Dissatisfaction subscale (EDI-BD; Garner, 1991; Appendix D) is a 9-item self-report measure that assesses women’s body dissatisfaction. Participants respond on a 6-point scale ranging from 1 (never true) to 6 (always true). A sample item is “I think my stomach is too big.” Higher scores indicate greater body dissatisfaction. The EDI-BD has demonstrated excellent internal consistency in past research, with alphas ranging from .89 to .91 (Brookings & Wilson, 1994; Tylka, 2004). Further, the EDI-BD has demonstrated convergent validity with other measures of body dissatisfaction, such as the Body Shape Questionnaire (r = .82; Garner, 1991). In the current study, the EDI-BD had excellent internal consistency, α = .90. To check whether participants were simply clicking through the questions without reading them, an additional item was added to the end of the EDI-2 asking participants to select the response “Usually.”
Covariates. The Sociocultural Attitudes Toward Appearance Scale-3 (SATAQ-3; Thompson, van den Berg, Roehrig, Guarda, & Heinberg, 2004; Appendix E) is a 30-item self-report measure of societal influences on body image. The SATAQ-3 consists of four subscales, one of which is the Internalization General subscale (SATAQ-IG). The
Eating Disorder Inventory (r = .57; Thompson et al., 2004). In the current study, the SATAQ-IG had excellent internal consistency, α = .94.
The Beck Depression Inventory-II (BDI-II; Beck, Steer, Ball, & Ranieri, 1996; Appendix F) is a 21-item self-report measure of depressive symptomatology. Participants respond on a 4-point scale ranging from 0 (absence of symptom; e.g., “I do not feel sad”) to 3 (severe presence of symptom; e.g., “I am so sad or unhappy that I can’t stand it”). Higher scores indicate greater severity of depressive symptoms. The BDI-II has
demonstrated excellent internal consistency in past research, α = 92 (Beck et al., 1996). Osman et al. (1997) also demonstrated adequate convergent validity between the BDI-II and other measures of depression (r = .77) and anxiety (r = .71). In the current study, the BDI-II had excellent internal consistency, α = .93. To check whether participants were simply clicking through the questions without reading them, an additional item was added to the end of the BDI-II asking participants to select the response “0.”
The Marlowe-Crowne Social Desirability Scale Form C (MCSDS-C; Reynolds, 1982; Appendix G) is a 13-item self-report measure of the tendency to respond to test items in a socially desirable manner. Participants indicate whether items are true or false for them personally. A sample item is “It is sometimes hard for me to go on with my work if I am not encouraged.” Higher scores indicate greater socially desirable responding. The MCSDS-C has demonstrated adequate internal consistency in past research (rKR-20 = .76; Reynolds, 1982). The MCSDS-C also has demonstrated convergent validity with other measures of social desirability, including the Edwards Social
conducted with and without the MCSDS-C. The MCSDS-C was excluded from final analyses because it did not significantly contribute to the regression models.
The Rosenberg Self-Esteem Scale (RSES; Rosenberg, 1965; 1979; Appendix H) is a 10-item self-report measure of global trait self-esteem. Participants respond on a 4-point scale ranging from 0 (strongly disagree) to 3 (strongly agree). A sample item is “I feel that I have a number of good qualities.” Higher scores indicate greater self-esteem. The RSES has demonstrated excellent internal consistency, α = .92 (Rosenberg, 1979). The RSES also has demonstrated convergent validity with other measures of self-esteem, including the Coopersmith Self-Esteem Inventory (r = .55; Demo, 1985). In the current study, the RSES had good internal consistency, α = .89. To check whether participants were simply clicking through the questions, an additional item was added to the end of the RSES asking participants to select the response “disagree.”
A demographic questionnaire (Appendix I) was used to obtain general demographic information, such as age and total years of university education. This questionnaire also asked for weight and height information to determine each
participant’s BMI. Body mass index was calculated by dividing each participant’s weight (in kilograms) by her height (in metres squared). Participants who did not provide weight and height information were omitted from the analyses.
Procedure
Study 1 was advertised on the Psychology Participant Pool as a study examining “Individual Differences and Perceptions of People” (see Appendix J for pool
participation, prior to the administration of the measures. Participants were encouraged to complete the study in a quiet area free from distractions. Consenting participants were directed to electronic forms of the questionnaires used in the study. The presentation order of the measures was as follows: OPTS, BDI-II, OBCSS, MCSDS-C, DPAC, RSES, SATAQ-IG, EDI and the demographic questionnaire. This order of presentation
alternated body image variables with non-appearance related variables. Upon completion of the measures, participants were directed to a debriefing page (Appendix L), which explained the purpose of the study and thanked them for their time and contribution. Finally, participants received a 0.5% bonus credit toward an eligible psychology course of their choice.
Study 1: Results
Approach to Data Analysis
All analyses were performed using SPSS for Mac (Version 25.0). Missing values and reliability analyses were conducted. Assumptions of multiple regression were assessed, followed by descriptive analyses. Finally, all of the hypotheses were tested using a series of multiple regression analyses using the Hayes (2017) PROCESS macro for SPSS. These steps were conducted first for the Caucasian-only sample, and then repeated for the full ethnically heterogeneous sample. Because of the limited sample size, separate analyses could not be conducted on women of non-Caucasian ethnicities.
Main Analyses for Caucasian-Only Sample
the analyses. Participants who failed more than one of the validity indicators (i.e., failed to
select the correct response on an item added to the EDI-2, BDI-II, and RSES; n = 3) were
not included in the analyses. After removing these three participants, as well as three
identified outliers (see Assumptions section below), the total Caucasian sample size for this study was 301 participants.
A missing values analysis was conducted on data from valid responders to assess for patterns of missingness. Seventy-nine percent (n = 240) of participants provided complete data. The percentage of missing values for all measure items ranged from 0 to 3.0%. Finally, less than 1% of all possible values were missing. Little’s MCAR test was not significant, χ2 (3922) = 3915.63, p = .526, indicating that the data were missing completely at random. This supported the use of imputation as an appropriate method of managing the missing data (Schafer & Graham, 2002). Expectation maximization was used to replace missing values, given the small amount of missing data (Tabachnick & Fidell, 2007).
Assumptions of multiple regression. The assumptions for multiple regression were examined according to the procedures outlined by Field (2009) and Tabachnick and Fidell (2007). The assumption of the absence of multicollinearity was assessed by
2009). Further, data were collected cross-sectionally, and there was no relationship between participants (Field, 2009). Accordingly, independence of errors was assumed. Next, the assumptions of normally distributed errors, homoscedasticity, and linearity were assessed. For each regression, the scatterplot of standardized residual versus
standardized predicted outcome appeared as a cloud, with an even concentration of scores around the centre. Furthermore, the scatterplot did not appear to have a wave or funnel pattern. Thus, linearity and homoscedasticity were assumed. Additionally, the histograms of standardized residuals approximated the normal curve, and the Shapiro-Wilk’s statistic for the standardized residuals was not significant, SW(301) = .993, p = .180. Thus,
Table 1
Zero-Order Correlations Between All Variables for Caucasian Sample (N = 301).
Variables 1 2 3 4 5 6 7 8 9 10
1. MCSDS-C -
2. RSES .21** -
3. BDI-II -.23** -.69** -
4. BMI .01 -.07 .04 -
5. SATAQ-IG -.25** -.33** .29** .07 -
6. OPTSneg -.12* -.04 .05 -.01 .15* -
7. APTSneg -.04 -.00 .07 -.03 .05 .26** -
8. OBCSS -.29** -.34** .25** .13* .58** .14* -.01 -
9. DPACS -.29** -.19** .23** .10 .50** .23** .09 .36** -
10. EDI-BD -.21** -.52** .49** .29** .53** .11 .03 .50** .32** -
Tabachnick and Fidell (2007) recommend assessing univariate normality for each
predictor. Based on the SW statistic, six predictors were not normally distributed. Thus, a
transformation was applied to each predictor. However, these transformations did not
reduce the SW statistics to non-significance, nor did they significantly change the results
of the final regression model (i.e., variables included in the final model, R2, regression
coefficients, significance values, etc.). Because the assumptions of homoscedasticity,
linearity, and normally distributed errors had been satisfied, and because predictor
variables are not assumed to be normally distributed in multiple regression, the
non-transformed predictor variables were used in the main analyses (Tabachnick & Fidell,
2007).
Finally, the data were examined for univariate outliers, residual outliers,
multivariate outliers, and influential cases. Three univariate outliers were identified (2 on
BDI and 1 on OPTSneg), and were replaced with the next closest value in the dataset
(Tabachnick & Fidell, 2007). Residual outliers were identified using standardized
residual values, and multivariate outliers were identified using both Mahalanobis distance
and leverage values. Three multivariate outliers were removed from all analyses.
Influential cases were examined using both Cook’s distance and DFFITS values. After
removing outliers in all regression analyses, no influential cases were identified.
Structure coefficients were examined for all variables included in the final
regressions (Courville & Thompson, 2001). The directional signs for all significant
regression coefficients were the same as those for the corresponding structure
model also were significantly correlated with the predicted outcome. Thus, no suppressor
variables were identified in the final models presented below.
Means and standard deviations for all variables are presented in Table 2. To
ensure specificity of the measure of fat stereotypes, the mean of the estimated
percentages of obese persons who possess negative traits (OPTSneg) was compared to
the mean of the estimated percentages of average-weight persons who possess the same
negative traits (APTSneg). A paired samples t-test found that participants estimated
significantly greater percentages of obese persons possessing the negative traits than they
did for average-weight persons possessing the same negative traits, t(300) = 12.96, p <
.001. Cohen’s d for this difference was .75, indicating a large effect.
Moderation analysis for body dissatisfaction. The first multiple regression
assessed Hypothesis 1, with body dissatisfaction as the criterion variable. The PROCESS
macro for Model 1 was used, as it examines potential moderation effects (Hayes, 2012).
In this regression, self-esteem, depressive symptomatology, and BMI were included as
significant covariates. Additionally, internalization of thin ideals was included to ensure
that any observed effects of endorsed fat stereotypes and body surveillance on body
dissatisfaction occurred above and beyond the effect of thin ideals. Fat stereotype
endorsement was included as the independent variable, and body surveillance was
entered as the moderator variable. The interaction term between fat stereotype
endorsement and body surveillance was then computed using the PROCESS macro. To
aid interpretation, the predictor and moderator variables were centered prior to
Table 2
Descriptive Statistics for Caucasian Sample (N = 301)
Variable N Range M SD Cronbach’s α
MCSDS-C 301 0.00 – 12.00 5.47 2.77 .6771
RSES 301 4.00 – 30.00 19.58 5.23 .890
BDI-II 301 0.00 – 49.00 13.86 10.49 .933
BMI 301 18.5 – 24.9 21.71 1.68 -
SATAQ-IG 301 1.00 – 5.00 3.13 1.00 .940
OPTSneg 301 17.00 – 91.00 58.37 12.63 .839
APTSneg 301 14.30 – 76.32 47.83 10.35 .837
OBCSS 301 1.50 – 7.00 4.72 1.06 .835
DPACS 301 1.00 – 5.00 2.86 1.05 .947
EDI-BD 301 9.00 – 54.00 30.87 9.45 .895
Note: 1 denotes a KR-20 value of internal consistency. MCSDS-C = Marlowe Crowne Social Desirability Scale Form C; RSES = Rosenberg Self-Esteem Scale; BDI-II = Beck Depression Inventory-II; BMI = Body Mass Index; SATAQ-IG = Sociocultural Attitudes Towards Appearance Scale-3 Internalization General subscale; OPTSneg = Obese
Table 3 provides a summary of the final model. The model was significant, F(7,
293) = 46.22, p < .001, accounting for 52.47% of the variance in body dissatisfaction. All
covariates significantly contributed to the model (all ps < .003). As expected,
endorsement of fat stereotypes did not significantly contribute to the model, b = 0.02,
t(300) = 0.75, p = .457. However, body surveillance was a significant predictor, b = 1.64,
t(300) = 3.59, p = <.001. Finally, adding the interaction term significantly improved the
prediction of body dissatisfaction, Fchange (1, 293) = 4.85, p = .028, accounting for an
additional 0.79% of the variance. As predicted, the interaction between fat stereotype
endorsement and body surveillance significantly contributed to the model, b = -.05,
t(300) = -2.21, p = .028. Because the interaction was significant, the specific effect of fat
stereotypes on body dissatisfaction at varying levels of body surveillance was examined
(see Table 4). Greater endorsement of fat stereotypes was significantly related to greater
body dissatisfaction in women who reported lower levels of body surveillance (1 SD
below the mean), t(300) = 1.97, p = .049, 95% CI [.001, .156]. In contrast, endorsement
of fat stereotypes was not significantly related to body dissatisfaction in participants who
reported higher levels of body surveillance (1 SD above the mean), t(300) = -1.03, p =
.302, 95% CI [-.108, .034]. Accordingly, Hypothesis 1 was partially supported. The
interaction between endorsed fat stereotypes and body surveillance significantly
predicted body dissatisfaction (see Figure 2). At lower levels of body surveillance,
greater fat stereotyping predicted higher body dissatisfaction. Contrary to the hypothesis,
however, fat stereotyping did not significantly predict body dissatisfaction at higher
levels of body surveillance. Notably, body surveillance levels at 1 SD above and below
Table 3
Moderation Model Summary Predicting Body Dissatisfaction for Caucasian Sample (N =
301)
95% CI
R R2 Variables Entered b SE b t p-value Lower Limit
Upper Limit
.724 .525 Constant 1.80 6.40 0.28 .778 -10.79 14.39 RSES -0.32 0.10 -3.11 .002 -0.53 -0.12 BDI-II 0.22 0.05 4.36 <.001 0.12 0.32 BMI 1.28 0.23 5.60 <.001 0.83 1.73 SATAQ-IG 2.52 0.48 5.30 <.001 1.58 3.46 OPTSneg 0.02 0.03 0.75 0.457 -0.03 0.07 OBCSS 1.64 0.46 3.59 <.001 0.74 2.54 OPTSnegxOBCSS -0.05 0.03 -2.20 .028 -0.10 -0.01
Table 4
Effect of Fat Stereotypes on Body Dissatisfaction at Varying Levels of Body Surveillance
for Caucasian Sample (N = 301)
95% CI Body
Surveillance Effect SE t p-value
Lower Limit
Upper Limit
1.06 .08 .04 1.97 .049 .001 .156
0.00 .02 .03 0.75 .457 -.033 .074
-1.06 -.04 .04 -1.03 .302 -.108 .034
Figure 2. The relationship between fat stereotype endorsement and body dissatisfaction at
lower and higher levels of body surveillance (N = 301). Fat Stereotype Endorsement 25
26 27 28 29 30 31 32 33 34
Low High
Bo
dy
D
issa
tisf
act
io
n
